Microsoft Threat Analysis Center General Manager Clint Watts says his team logged hundreds of instances of Russia, Iran and China using artificial intelligence during the last 14 months and encountered some unexpected results.

Mr. Watts told the recent Billington CyberSecurity Summit in Washington that deepfakes do not work on large audiences as predicted.

“The story a year ago was deepfakes in the election, and candidates being deepfaked — audiences have been remarkably brilliant about detecting deepfakes in crowds,” Mr. Watts said. Instead, he said, it is when people are alone that they tend to fall for the manipulation. 

Mr. Watts also warned of deepfake audio tricks on the horizon. “If it’s done right and used in the last factor, which is at a specific condition or time — times of crisis, conflict, competition — people tend to fall for things they wouldn’t normally fall for,” he said.

A top NSA official recently gave Threat Status insight into the agency’s efforts to partner with private companies to thwart digital disruptions to systems tied to the election. NSA Cybersecurity Collaboration Center Director Kristina Walter said in an exclusive interview that the agency works “very closely” with its Election Security Group, which is focused squarely on foreign threats to elections.

“If there’s an attempt for a malicious actor to abuse a U.S. platform to target elections, we can share that information through the collaboration center from the Election Security Group with that partner so that they can disrupt the activity,” Ms. Walter told Threat Status on the sidelines of the recent Billington CyberSecurity Summit in Washington.

The center has enlisted more than 1,000 cybersecurity companies to help foil hackers’ plans since its launch approximately four years ago. Ms. Walter would not identify which specific companies the agency has worked with, but pointed generally to internet service providers, cloud providers, managed service providers and cybersecurity companies as capable of defending elections.

While the NSA is reserved about its work with the private sector, cybersecurity pros in the private sector are less guarded. Mr. Lovelace, who interviewed Ms. Walter, offers this deep dive on the situation.

U.S. military officials are bracing for a major breakthrough that could accelerate the arrival of revolutionary quantum computers, and are readying new initiatives to validate discoveries and test the usefulness of the machines.

Quantum computers present risks and opportunities for national security, such as through “cryptanalytically relevant quantum computers” capable of breaking encryption that secures defense secrets and financial transactions.

John Burke, the Defense Department’s principal director of quantum science, discussed the U.S. military’s preparations in remarks at this week’s Quantum World Congress in Virginia, asserting that the Pentagon’s top priority for quantum computing is to understand its uses and prevent the machines from falling into adversaries’ hands.

“There will be a time — and that time might come very soon so we’re going to be open-minded to that — that quantum computing will be a very important application for useful applications for the DoD,” said Mr. Burke. “That will change things dramatically if we have that evidence of utility.”

Joe Saunders from RunSafe Security, Marke Eadie from Coder, Tom Hofmann from Flashpoint, Jon France from ISC2, and Tim LeMaster from Lookout are among the guests Threat Status interviewed at last week’s Billington CyberSecurity Summit.

The Threat Status weekly podcast features industry leaders from the floor of the gathering, where hundreds of private cybersecurity firms spent the week mingling with U.S. intelligence and defense officials.

The podcast features discussions on the future of AI in intelligence gathering, foreign adversary hacking of U.S. systems, America’s own offensive cyber operations, and the possibility of a “Cyber Doomsday” attack before the November U.S. presidential election.

Listen here or wherever you get your podcasts.

Ireland’s Data Protection Commission (DPC) said Thursday that it has opened an inquiry into Google’s Pathway’s Language Model 2, also known as PaLM2, as part of a wider effort by other national watchdogs within the 27-country EU to scrutinize how AI systems handle personal data.

The Associated Press notes that Google’s European headquarters are based in Dublin, so the Irish watchdog acts as the company’s lead regulator for the EU’s privacy rulebook, known as the General Data Protection Regulation, or GDPR.

Ireland’s DPC said in a statement that it wants to know if Google has assessed whether PaLM2’s data processing would likely result in a “high risk to the rights and freedoms of individuals” in the EU.

Large language models like PaLM2 are vast troves of data that act as building blocks for artificial intelligence systems. Google uses PaLM2 to power a range of generative AI services, including email summarizing.

Tech Crunch reported Thursday that Google spokesman Jay Stoll said the company “take[s] seriously our obligations under the GDPR and will work constructively with the DPC to answer their questions.”

• Sept. 18 — Aspen Cyber Summit, The Aspen Institute

• Sept. 18-19 — Chief Data & Analytics Officers: Government 2024, Corinium Intelligence

• Sept. 20 — Confronting the Axis of Upheaval, Center for a New American Security

• Oct. 5-8 — 2024 Threat Conference, The Cipher Brief

• Oct. 21-24 — Gartner IT Symposium/Xpo, Garter Conferences

• Nov. 21 — Competition Policy 2024: Urgent Questions Emerging within Digital Markets, Chatham House